山裡撿到一撮毛，一坨糞便，過去送驗常常卡在同一個問題：粒線體基因裡混進了核基因體裡的偽基因片段（numts），結果比對出來的物種，可能跟真正留下痕跡的動物對不上號。這有點像拿到一張被貼錯標籤的罐頭，成分表看起來沒錯，裡面裝的卻是另一種東西。
農業部林業及自然保育署的研究團隊，在期刊ZooKeys上發表了台灣第一份非飛行性陸棲哺乳類粒線體COI DNA條形碼參考庫，刊在1293卷，45到77頁。這份資料庫系統性覆蓋了台灣主要的陸棲哺乳動物，等於替往後每一撮毛髮、每一坨糞便，建了一本可以對照的身分底冊。
偽基因的陷阱
numts問題聽起來技術，講白了就是DNA比對會騙人。粒線體基因體在演化過程中，片段會被複製貼進細胞核基因體，形成一段長得很像、但其實不是粒線體基因的「仿冒序列」。鑑定時若誤抓到這段仿冒序列當樣本，比對結果會指向錯誤物種，或者讓同一隻動物看起來像兩種不同生物。這份參考庫把這層干擾處理掉，等於先把罐頭標籤對準了再上架。
這不只是學術上的講究。野生動物走私查緝靠的就是DNA快速鑑定——查獲的肉乾、骨頭、毛皮,送驗時如果連物種都認錯，整個案子的證據力就垮了。保育監測也一樣,紅外線相機拍不到、但毛髮和糞便可以採樣的地方,DNA條形碼就是唯一的追蹤手段。
黑熊樣本的特殊之處
研究特別點名了台灣黑熊樣本的特殊性,也對鼬獾屬（Melogale）的分類地位提出具體建議——鼬獾在台灣山林裡並不罕見,但牠在分類學上的位置,一直有討論空間。這份參考庫把這類懸而未決的問題攤開,用分子證據重新梳理。
台灣黑熊（Ursus thibetanus formosanus）本身已經不是陌生面孔。2026年3月,林業及自然保育署公布全島首次族群估算,約1,200隻,且從2011年以來持續向淺山平地擴展——族群數字漂亮,人熊相遇的次數也跟著往上走。4月,花蓮卓溪鄉完成台灣首次母子熊聯合野放,布農族部落領袖參與了野放儀式;同一個月,台東卑南鄉的低海拔森林相機也拍到了黑熊身影,那是過去少見的海拔高度。一邊是族群回升的好消息,一邊是棲地被壓縮到連熊都得往平地搬家的壞消息,兩條線同時發生。
一座島,十五萬種可能
台灣3.6萬平方公里的土地上,已記錄的生物種類超過5.9萬種,這個密度常被拿來跟同緯度國家比較。在這樣擁擠的生態舞台上，每一種動物都需要一套可靠的身分辨識方法，才談得上精準保育。COI條形碼參考庫做的事說穿了很樸素：讓以後的研究者、查緝人員、保育工作者，拿到一撮毛、一塊骨頭，不用再猜。
刊物頁數是45到77頁，數字不長，但這本「身分底冊」接下來會被引用多少次，才是真正的考題。
— 鍾雅筑
A Strand of Bear Hair Just Got a DNA Rulebook
Pick up a tuft of hair on a mountain trail, or a pile of scat, and identification used to hit the same snag: mitochondrial DNA readings got contaminated by nuclear copies of the same genes — fragments called numts — scrambling which species the sample actually belonged to. It’s a bit like a canned good with the right-looking label pasted onto the wrong contents.
A research team under Taiwan’s Forestry and Nature Conservation Agency published the island’s first mitochondrial COI DNA barcode reference library for non-volant terrestrial mammals, in ZooKeys volume 1293, pages 45-77. The library systematically covers Taiwan’s major land mammal species — essentially a lookup table for every future hair sample or scat pellet found in the field.
The numts trap
The numts problem sounds technical, but the practical effect is simple: DNA matching can lie. Over evolutionary time, fragments of mitochondrial genes get copied into the nuclear genome, producing near-identical “counterfeit” sequences. Grab the wrong one during testing and the result points to the wrong species — or makes one animal look like two. The new reference library filters this interference out before it reaches the lab bench.
This isn’t academic tidiness. Wildlife trafficking enforcement depends on rapid DNA identification — jerky, bones, pelts seized at a checkpoint are useless as evidence if the species call is wrong. Conservation monitoring runs into the same wall wherever camera traps can’t reach but hair and scat can be collected. Barcoding is the only tracking method left.
What’s special about the bear samples
The study flags particular quirks in its Formosan black bear samples and offers concrete taxonomic recommendations for the ferret-badger genus Melogale — a genus common enough in Taiwan’s forests, but whose classification has long been debated. The reference library lays that question out with molecular evidence.
The Formosan black bear (Ursus thibetanus formosanus) was already a familiar face. In March 2026, the Forestry and Nature Conservation Agency released the island’s first nationwide population estimate — roughly 1,200 — noting continued expansion from high mountains into lower-elevation areas since 2011. In April, Hualien’s Zhuoxi Township carried out Taiwan’s first joint release of a rehabilitated mother bear and cub, with Bunun community leaders taking part in the release. The same month, camera traps in Taitung’s Beinan Township caught a bear at an elevation rarely recorded before. Population recovery and habitat squeeze are running on the same timeline.
One island, tens of thousands of species
Taiwan’s 36,000 square kilometers host more than 59,000 recorded species, a density regularly compared against countries of similar latitude. On a stage that crowded, reliable species identification isn’t a luxury — it’s the baseline for any conservation work worth doing. What the COI barcode library does is unglamorous: let the next researcher, customs officer, or field biologist holding a tuft of fur skip the guesswork.
Thirty-three journal pages, in the end. Whether they get cited matters more than their length.
— 鍾雅筑
延伸閱讀
https://justfly.idv.tw/s/yU8WnET